How Embeddings Are Created

1. What is the first step in the embedding creation process?

Tokenization Contextual encoding Pooling Vectorization

2. Why is it important to use the same embedding model for all your text?

Different models use unique dimensions Models have differing input requirements Comparing vectors from different models is meaningful Embeddings from different models live in the same space

3. What does each number in an embedding vector represent?

A token in the text A fixed feature of meaning A model parameter A piece of contextual information

4. Which embedding model runs locally and avoids data egress in enterprise applications?

nomic-embed-text Azure OpenAI Google Embeddings Hugging Face Transformers

Continue: Similarity and Distance → Review this lesson Retake quiz